CN117973859A

CN117973859A - Risk prevention and control method, device, equipment and medium for main distribution network

Info

Publication number: CN117973859A
Application number: CN202410123032.4A
Authority: CN
Inventors: 梁志峰; 王鹤; 隋凌峰; 边竞; 霍雪松; 陈文进; 杜云龙; 柴赟; 于若英; 陈菁伟; 杨晓雷
Original assignee: State Grid Corp of China SGCC; State Grid Zhejiang Electric Power Co Ltd; China Electric Power Research Institute Co Ltd CEPRI; State Grid Jiangsu Electric Power Co Ltd; Northeast Dianli University
Current assignee: State Grid Corp of China SGCC; State Grid Zhejiang Electric Power Co Ltd; China Electric Power Research Institute Co Ltd CEPRI; State Grid Jiangsu Electric Power Co Ltd; Northeast Electric Power University
Priority date: 2024-01-29
Filing date: 2024-01-29
Publication date: 2024-05-03

Abstract

The invention discloses a risk prevention and control method, device, equipment and medium for a main distribution network. The method comprises the steps of obtaining wind-light output prediction data and wind-light output actual measurement data within a preset time period; determining a wind and light output scene set according to wind and light output prediction data, wind and light output actual measurement data, pre-added random noise and a pre-trained WGAN-GP model; scene reduction is carried out on the scene set of the wind and light output according to a preset clustering algorithm, and a target scene set of the wind and light output after scene reduction is obtained; performing risk prevention and control on a main power distribution network in each wind-light output scene in a target wind-light output scene set based on a pre-constructed risk prevention and control model; according to the technical scheme, the condition that the traditional control unit is only used for adjusting can be improved, and the running risk of the main distribution network can be effectively reduced.

Description

Risk prevention and control method, device, equipment and medium for main distribution network

Technical Field

The invention relates to the technical field of safety of distribution networks, in particular to a risk prevention and control method, device, equipment and storage medium for a main distribution network.

Background

In recent years, photovoltaic power generation technology has gradually tended to be mature and perfect, grid-connected capacity of centralized and distributed photovoltaic is gradually increased, the duty ratio is continuously improved, meanwhile, as the top distributed photovoltaic is pushed in a large scale, the distributed photovoltaic is continuously developed in a large scale, and new requirements are also provided for strengthening the power grid dispatching management level. How to acquire the running state information of the distributed new energy source in real time, analyze the running risk points of the distributed new energy source, adopt a dispatching control strategy to eliminate the running risk of the power grid, improve the running stability and reliability of the main power distribution network, and is a problem to be solved by dispatching decision-making personnel. At present, distributed new energy is rapidly developed in recent years, the capacity of a distributed installation in some areas exceeds 1000 kilowatts, and the installation in other areas exceeds 150 kilowatts. Because the grid-connected voltage level is low, the operation information of the grid-connected voltage level is not completely mastered by scheduling, and great potential safety hazards are brought to the operation of the power grid, and the three aspects of safety, balance and digestion are mainly embodied, and the method is as follows:

The safety operation risk of the power grid is increased, the distributed new energy source is developed in a large scale, and the main power distribution network is gradually developed from a passive network to an active network. The photovoltaic power generation capacity of part of the transformer areas exceeds the user consumption capacity of the transformer areas, and a large amount of electric energy is sent up, so that the transformer areas and low-voltage lines are overloaded, even 110 (35) kilovolts, 220 kilovolts transformer substations or lines are reversely overloaded, and the safe and stable operation of the power grid is endangered. And when the active power of the distributed photovoltaic output is larger, the terminal voltage of a distribution line and a transformer area is easy to raise, and the voltage regulation difficulty of a power grid is increased. In addition, a large amount of distributed new energy is accessed, so that the activation of a low-frequency low-voltage cut feeder line is aggravated, the risk of insufficient cut load exists, and the security of a three-defense line of a system is affected.

At present, active main distribution network risk assessment mainly researches influences of various uncertainty factors, and mainly comprises electric vehicle charging and discharging influences when a plurality of types of electric vehicles are connected to the main distribution network and influences on new energy sources such as photovoltaic and the like, which are connected to the main distribution network. However, such risk assessment methods, while able to affect the probabilistic nature of such uncertainty factors and the extent to which it affects, do not give a reasonable measure to reduce the operational risk of the main distribution network.

Disclosure of Invention

In view of the above, the invention provides a risk prevention and control method, device, equipment and storage medium for a main distribution network, which can improve the condition of traditional regulation only by a controllable unit and can effectively reduce the running risk of the main distribution network.

According to an aspect of the present invention, an embodiment of the present invention provides a risk prevention and control method for a main power distribution network, where the method includes:

acquiring wind-light output prediction data and wind-light output actual measurement data within a preset time period;

Determining a wind and light output scene set according to the wind and light output prediction data, the wind and light output actual measurement data, the pre-added random noise and a pre-trained generation countermeasure network WGAN-GP model with gradient penalty;

performing scene reduction on the scene set of the wind and light output according to a preset clustering algorithm to obtain a scene reduced target scene set of the wind and light output;

Performing risk prevention and control on a main power distribution network in each wind-light output scene in the target wind-light output scene set based on a pre-constructed risk prevention and control model; the risk prevention and control model is a main power distribution network risk prevention and control model which aims at minimum comprehensive operation risk indexes and minimum comprehensive operation cost.

In an embodiment, the determining the wind-light-output scene set according to the wind-light-output prediction data, the wind-light-output actual measurement data, the pre-added random noise and the pre-trained WGAN-GP model further includes:

Inputting the wind-light output prediction data and at least two groups of random noise into a generator in a pre-trained WGAN-GP model to obtain generation data, and inputting the generation data into a discriminator in the pre-trained WGAN-GP model; wherein the random noise is in accordance with normal distribution;

inputting the wind-solar output actual measurement data into a discriminator in a pre-trained WGAN-GP model;

And outputting and obtaining at least two wind and light output scenes from the discriminator, and forming a wind and light output scene set by the at least two wind and light output scenes.

In one embodiment, the WGAN-GP model training process includes:

Acquiring historical wind-solar power output data of a preset area as a sample set; the sample set comprises a historical wind-light output predicted value and a historical wind-light output actual measurement value;

splicing the pre-added random noise and the historical wind-light output predicted value, and inputting the spliced random noise and the historical wind-light output predicted value into a generator of the WGAN-GP model to output a generated sample;

Inputting the history wind-solar output actual measurement value into a discriminator of the WGAN-GP model, inputting the generated sample into the discriminator, and outputting a discrimination value for the history wind-solar output actual measurement value and the generated sample;

and adding a gradient penalty term in the loss function of the discriminator, wherein the gradient penalty term has a weight coefficient, respectively updating the weight coefficient by adopting a preset deep learning optimization algorithm, and returning to the step of acquiring the historical wind-solar output data of the preset area as a sample set until the difference between the generated sample output by the discriminator and the actual measurement value of the historical wind-solar output is minimum.

In one embodiment, the objective function in the arbiter is formulated as: wherein E (·) represents the expected value; d (G (z)) is a probability that the generated data G (z) is discriminated as true in the discriminator; d (x) represents the probability that the true data x is discriminated as true in the discriminator; the distribution of the real data x is x-p _data(x); z represents the random noise, wherein the distribution of the random noise is z-P _z;P_X which is the real distribution of wind and light data x; x' =εx+ (1- ε) G (z); epsilon represents a random number; λ represents a weight coefficient of the gradient penalty term; the expression of the norm is a mathematical calculation method; denoted hamiltonian.

In an embodiment, the performing scene reduction on the scene set according to a preset clustering algorithm to obtain a target scene set of wind and light output after scene reduction, further includes:

Selecting a preset number of wind-light output scenes from the wind-light output scene set as an initial clustering center;

Determining a first distance from other wind-light output scenes in the wind-light output scene set to the initial clustering center by adopting a preset similarity measurement method, and dividing the other wind-light output scenes to the closest clustering center according to the first distance to form a first clustering center; wherein the first distance characterizes similarity;

For each first clustering center, respectively calculating a second distance of a wind-light output scene in each first clustering center, updating the first clustering centers according to the second distance to obtain a new clustering center, and returning to the step of dividing the other wind-light output scenes to the closest clustering center according to the first distance to form a first clustering center until the first clustering center is not changed any more or reaches a preset iteration number to obtain a target clustering result; wherein the second distance characterizes a total distance of the wind-light output scenes in the first cluster center;

determining a reduced target wind-light output scene set according to the target clustering result;

Wherein, the target clustering result comprises: the number of samples or sample weights included in each cluster; and the scene probability sum of all the wind-light power scenes is 1.

In an embodiment, the building of the risk prevention and control model further includes:

acquiring power flow parameter data of the power distribution network in each wind-light power generation scene during operation;

Determining at least two power distribution network safety risk indexes corresponding to the wind-light output scenes within a preset time granularity based on scene probabilities respectively corresponding to the wind-light output scenes and the tide parameter data aiming at each wind-light output scene in the target wind-light output scene set;

constructing a risk prevention and control model corresponding to the wind-light output scene according to the safety risk indexes of each power distribution network; wherein, the distribution network security risk index at least includes: node voltage out-of-limit risk indicators, line power out-of-limit risk indicators and system load loss risk indicators.

In an embodiment, the power flow parameter data includes: under the conditions of the voltage amplitude of the node i and the upper limit and the lower limit of the voltage amplitude of the node i in the power distribution network, the node voltage out-of-limit risk index is expressed as follows: Wherein ρ (E _k) is the scene probability of the power distribution network node system state E _k; n _D is the total number of nodes of the power distribution network node system; v _i is the voltage amplitude of node i; /(I) V _i is the upper limit and the lower limit of the voltage amplitude of the node i respectively;

The tide parameter data comprises: active power sum of line l transmission In the case of the maximum active power allowed to be transmitted by the line l, the line power out-of-limit risk indicator is expressed as: Wherein ρ (E _k) is the scene probability of the system state E _k; n _L is the total number of system lines; p _l is the active power transmitted by line l; /(I) Maximum active power allowed to be transmitted for line l;

the tide parameter data comprises: under the conditions of the active load quantity of the node i and the active load value of the node i, the system load loss risk index is expressed as follows by a formula: Wherein ρ (E _k) is the scene probability of the system state E _k; p _cut,i is the active load amount of node i; p _d,i is the active load value of node i.

In one embodiment, the pre-constructed risk prevention and control model is formulated as:

Wherein, Expressed as minimum integrated operational risk; /(I)Expressed as minimum integrated operating cost; r _V is a node voltage out-of-limit risk index; r _L is a line power out-of-limit risk index; r _C is a system load loss risk index; wherein ρ _t is the purchase price of electricity in different time periods; p _loss (t) is the net loss; s _Gi (t) is the cost of the controllable unit; z _Pi (t) is the interrupt cost; n _G is expressed as the number of controllable units; n _l represents the number of user interruptible loads;

The constraint conditions of the risk prevention and control model comprise: demand response constraints, energy storage system ESS constraints, fan constraints, photovoltaic constraints, generator constraints, reactive compensation device SVC constraints, capacitor CB constraints, node voltage constraints, and branch current constraints.

In one embodiment, the demand response constraint is formulated as: And Wherein/>The variable quantity generated by the response of the demand side at the time t of the n node is obtained; /(I)The load quantity before the response of the demand side at the moment t of the n node is; /(I)And/>Respectively generating upper and lower limits of variation for the response of the demand side at the time t of the n node; kappa is the allowable variable range of electricity consumption in one period of the user; t is expressed as the number of hours of a day, and the value is 24; n _b represents the total number of nodes;

The charge and discharge state constraint in the ESS constraint of the energy storage system is expressed as: Wherein/> And/>Respectively charging and discharging states of the energy storage j at the moment t, and indicating 3 states of charging, discharging and non-charging and non-discharging when the constraint limit is within 1; n ^ES is the number of energy storage devices in the main distribution grid;

the fan constraint is expressed as: Wherein/> And/>Active power and reactive power sent by the fan j at the moment t are respectively; /(I)The maximum active power emitted by the fan j at the moment t; tan θ _WT is the fan power factor; n ^WT is the number of fans in the main power distribution network;

the photovoltaic constraint is expressed as: Wherein/> And/>Active power and reactive power sent by the fan j at the moment t are respectively; /(I)The maximum active power emitted by the fan j at the moment t; tan θ _PV is the photovoltaic power factor; n ^PV is the number of fans in the main power distribution network;

the generator constraint is expressed as: Wherein/> Active power generated by the generator j at the moment t; /(I)Maximum active power generated by the generator j at the time t; n ^MT is the number of generators in the main distribution network;

The SVC constraint of the reactive power compensation device is expressed as: Wherein/> The reactive compensation quantity of the continuous reactive compensation device j at the time t is obtained; /(I)And/>Reactive compensation upper and lower limits of the continuous reactive compensation device j; n ^SVC is the number of continuous reactive power compensation devices in the main power distribution network;

The capacitor CB constraint is formulated as: Wherein/> The reactive compensation quantity of the discrete reactive compensation device j at the time t is obtained; /(I)Switching the group number for the discrete reactive power compensation device j at the time t; reactive compensation quantity switched for a single group of discrete reactive compensation devices j; /(I) The maximum switching group number of the discrete reactive power compensation device j; n ^CB is the number of discrete reactive power compensation devices in the main power distribution network;

the node voltage constraint is expressed as: wherein V _n,t is the voltage of the node n at the time t; /(I) And/>Upper and lower limits allowed for the voltage on node n;

The branch current constraint is expressed as: Wherein/> And/>Upper and lower limits for the branch l to allow the flow of current; i _l,t is the value of the current flowing in the branch l at time t.

In an embodiment, the risk prevention and control of the main distribution network in each wind-light output scene in the target wind-light output scene set based on the pre-constructed risk prevention and control model includes:

solving the optimal power flow of the risk prevention and control model by adopting a preset particle swarm algorithm;

And adjusting energy storage and demand response of the main distribution network in each wind-light output scene according to the optimal power flow so as to perform risk prevention and control on the main distribution network.

In an embodiment, the solving the optimal power flow of the risk prevention and control model by adopting a preset particle swarm algorithm includes:

Randomly initializing the current position and the current speed of the initial group of particles; wherein the initial population of particles represents a possible solution; the position of each particle represents a set of solutions to the node voltage and generator output parameters in the power system, and the velocity of each particle represents the direction and velocity of the particle searching in the solution space;

Updating the current position according to the first historical optimal solution of the particles and the second historical optimal solution of the whole population to obtain an updated new position;

Calculating an output value of the risk prevention and control model according to the new position; the output value represents the fitness of the optimal power flow problem;

updating the individual optimal solution of each particle according to the fitness, and selecting the optimal solution as a group optimal solution;

Judging whether the group optimal solution meets a stopping condition, and if so, returning to the optimal solution; if not, returning to the step of updating the current position according to the first historical optimal solution of the particles and the second historical optimal solution of the whole population to obtain an updated new position, and continuing to update the position and the speed of the particles until the stopping condition is met; wherein the stop condition includes one of: the objective function value converges to a preset threshold value, the objective function value meets constraint conditions, and the preset iteration times are reached.

According to another aspect of the present invention, an embodiment of the present invention further provides a risk prevention and control device for a main power distribution network, where the device includes:

The acquisition module is used for acquiring wind-light output prediction data and wind-light output actual measurement data in a preset time period;

the determining module is used for determining a wind and light output scene set according to the wind and light output prediction data, the wind and light output actual measurement data, the pre-added random noise and a pre-trained WGAN-GP model;

the reduction module is used for carrying out scene reduction on the scene set of the wind and light output according to a preset clustering algorithm to obtain a target scene set of the wind and light output after scene reduction;

The prevention and control module is used for performing risk prevention and control on the main distribution network in each wind-light output scene in the target wind-light output scene set based on a pre-constructed risk prevention and control model; the minimum risk prevention and control model is a main power distribution network risk prevention and control model which aims at comprehensive operation risk indexes and minimum comprehensive operation cost.

According to another aspect of the present invention, an embodiment of the present invention further provides an electronic device, including:

At least one processor; and

A memory communicatively coupled to the at least one processor; wherein,

The memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the risk prevention and control method of the main distribution network according to any embodiment of the present invention.

According to another aspect of the present invention, an embodiment of the present invention further provides a computer readable storage medium, where computer instructions are stored, where the computer instructions are configured to enable a processor to implement the risk prevention and control method of the main distribution network according to any one of the embodiments of the present invention when executed.

According to the technical scheme, the wind-light output scene set is determined through the obtained wind-light output prediction data, the wind-light output actual measurement data, the pre-added random noise and the pre-trained WGAN-GP model, the wind-light output scene set is subjected to scene reduction according to a preset clustering algorithm, a target wind-light output scene set after scene reduction is obtained, scene reduction can be carried out, typical characteristics of wind-light output can be better described by the reduced scene, and the fluctuation of wind-light output is reflected; and performing risk prevention and control on the main power distribution network in each wind-light output scene in the target wind-light output scene set through a pre-constructed risk prevention and control model, so that the condition that the traditional wind-light output scene set is regulated only by a controllable unit can be improved, and the running risk of the main power distribution network can be effectively reduced.

It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the invention or to delineate the scope of the invention. Other features of the present invention will become apparent from the description that follows.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a flowchart of a risk prevention and control method for a main distribution network according to an embodiment of the present invention;

FIG. 2 is a basic structure of WGAN-GP model according to one embodiment of the present invention;

Fig. 3 is a flowchart of another risk prevention and control method for a main distribution network according to an embodiment of the present invention;

FIG. 4 is a graph of a reduced 5 exemplary wind power output scenario according to an embodiment of the present invention;

FIG. 5 is a graph of a scene of reduced 5 typical photovoltaic output according to an embodiment of the present invention;

FIG. 6 is a graph showing a trend of power variation of an interruptible load 24h according to an embodiment of the present invention;

FIG. 7 is a graph showing a trend of power variation of a load 24h with time shift according to an embodiment of the present invention;

FIG. 8 is a graph showing a trend of the power change of the energy storage device 24h according to an embodiment of the present invention;

fig. 9 is a topology structure diagram of risk prevention and control for a main distribution network according to an embodiment of the present invention;

fig. 10 is a block diagram of a risk prevention and control device for a main power distribution network according to an embodiment of the present invention;

fig. 11 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.

It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

In an embodiment, fig. 1 is a flowchart of a risk prevention and control method for a main power distribution network according to an embodiment of the present invention, where the method may be applicable to a situation when risk prevention and control is performed on a main power distribution network in an electric power internet of things, the method may be performed by a risk prevention and control device for the main power distribution network, and the risk prevention and control device for the main power distribution network may be implemented in a form of hardware and/or software, and the risk prevention and control device for the main power distribution network may be configured in an electronic device.

As shown in fig. 1, the risk prevention and control method for the main distribution network in this embodiment includes the following specific steps:

S110, wind and light output prediction data and wind and light output actual measurement data in a preset time period are obtained.

The preset time period may be understood as a history of a certain period of time. The wind-light output prediction data refer to power prediction data respectively corresponding to wind power and photovoltaic in a certain period of time in a certain region; the wind-light output actual measurement data refers to the actual measurement data of the power corresponding to wind power and photovoltaic respectively in a certain period of time in a certain region, namely the power data of the region tested in real time.

In this embodiment, power prediction data and power actual measurement data corresponding to wind power and photovoltaic in a certain period of time in a certain region can be obtained, and a corresponding wind-light output scene can be constructed through the power prediction data and the power actual measurement data. Exemplary are wind power generation prediction data, photovoltaic power generation prediction data, wind power generation actual measurement data, and photovoltaic power generation actual measurement data for each day of the year.

S120, determining a wind and light output scene set according to wind and light output prediction data, wind and light output actual measurement data, pre-added random noise and a pre-trained generation countermeasure network WGAN-GP model with gradient penalty.

Wherein, the pre-added random noise refers to the random noise added by people, and the random noise accords with normal distribution. Illustratively, n high-dimensional noises conforming to a normal distribution are added. The wind-light output scene set may be composed of one or more wind-light output scenes, and it may be understood that n wind-light output scenes constitute the wind-light output scene set.

In this embodiment, the generation countermeasure network (WASSERSTEIN GAN WITH GRADIENT PENALTY, WGAN-GP) model with gradient penalty is WGAN model introducing gradient penalty, WGAN-GP uses wasperstein distance to measure the distance between the generated data distribution and the actual data distribution. The gradient penalty term of function D is introduced in the defined domain to ensure that the discriminator function approximately satisfies the liplitz continuum.

In this embodiment, wind-light output prediction data and multiple sets of random noise can be input into a generator in a pre-trained WGAN-GP model to obtain generation data, the generation data is input into a discriminator in a pre-trained WGAN-GP model, wind-light output actual measurement data is input into a discriminator in a pre-trained WGAN-GP model, and multiple wind-light output scenes are output from the discriminator to form a wind-light output scene set.

In some embodiments, the training process of WGAN-GP model includes:

Splicing the pre-added random noise and the historical wind-light output predicted value, and inputting the spliced random noise and the historical wind-light output predicted value into a generator of a WGAN-GP model to output a generated sample;

Inputting the actual measurement value of the historical wind and light output into a WGAN-GP model discriminator, inputting a generated sample into the discriminator, and outputting the actual measurement value of the historical wind and light output and the discrimination value of the generated sample;

And adding a gradient penalty term in the loss function of the discriminator, wherein the gradient penalty term has a weight coefficient, respectively updating the weight coefficient by adopting a preset deep learning optimization algorithm, and returning to the step of acquiring the historical wind-light output data of the preset region as a sample set until the difference between the generated sample output by the discriminator and the actual measurement value of the historical wind-light output is minimum.

In this embodiment, the WGAN-GP model selects part of sampling points in the domain of the discriminant function D to perform gradient calculation, and the sampling points are distributed on the connection line between the generated sample and the real sample for each training. When the WGAN-GP model is trained, the predicted and measured samples are subjected to normalization processing, high-dimensional noise which accords with standard normal distribution is extracted, the noise and predicted values in the training samples are longitudinally spliced and then input into a generator, and the generator outputs generated samples. And (3) longitudinally splicing the predicted value (namely the generated sample in the graph) with the measured value, inputting the longitudinally spliced predicted value and the generated sample into the discriminator, outputting the discriminated value of the real sample and the generated sample by the discriminator, selecting a gradient punishment sampling point, calculating a loss function of the generator and the discriminator, respectively updating the network weights of the generator and the discriminator by adopting a root mean square propagation RMSprop optimization algorithm, repeating the whole process until the difference between the generated sample output by the discriminator and the measured value of the historical wind-solar output reaches the minimum, and ending training.

In this embodiment, the objective function in the arbiter is formulated as: Wherein E (·) represents the expected value; d (G (z)) is a probability that the generated data G (z) is judged to be true in the discriminator; d (x) represents the probability that the true data x is judged to be true in the discriminator; the distribution of the real data x is x-p _data(x); z represents random noise, wherein the distribution of the random noise is z-P _z;P_X which is the real distribution of wind and light data x; x' =εx+ (1- ε) G (z); epsilon represents a random number; λ represents the weight coefficient of the gradient penalty term; the expression of the norm is a mathematical calculation method; denoted hamiltonian.

Illustratively, to facilitate a better understanding of the basic structure of the WGAN-GP model, FIG. 2 is a schematic diagram of the basic structure of a WGAN-GP model according to one embodiment of the present invention. In this embodiment, by setting the objective function, a corresponding loss function can be obtained; in the subsequent training process, the generator GN and the discriminator DN are alternately trained, so that the generating capacity and the discriminating capacity are continuously enhanced, and the generator can learn the characteristics of the training sample space better. The wind-light output has rich characteristics, and the WGAN-GP characteristics are just suitable for extracting dynamic information of a wind-light output curve, so that the quality of a sample generated by wind-light output is higher.

S130, performing scene reduction on the scene set of the wind and light output according to a preset clustering algorithm to obtain a target scene set of the wind and light output after scene reduction.

The preset clustering algorithm can comprise a K-means clustering method and a K-means clustering method, wherein the K-means clustering method is a clustering algorithm based on similarity among samples. The scene probability refers to a scene probability corresponding to each wind-light output scene included in the reduced target wind-light output scene set, and the target wind-light output scene set may include at least two wind-light output scenes; it can be understood that each scene of wind and light output corresponds to a scene probability, and the sum of all scene probabilities is 1.

In this embodiment, a preset number of wind-light output scenes can be selected from the wind-light output scene set as an initial clustering center, a preset similarity measurement method is adopted to determine distances from other wind-light output scenes in the wind-light output scene set to the initial clustering center, scene division is performed according to the distances to form a clustering center, the distances represent similarity, the clustering center is updated to obtain a new clustering center until the clustering center is not changed any more or a preset iteration number is reached, a target clustering result is obtained, and a reduced scene set is determined according to the target clustering result; in some embodiments, the wind and light scenes can be clustered based on a Kmeans algorithm, so that the reduction of a large-scale scene is realized, the reduction is carried out to a target scene after the reduction, and finally the probability of each scene is obtained and multiplied and summed with each corresponding scene to obtain uncertainty output; the present embodiment is not limited herein.

And S140, performing risk prevention and control on the main power distribution network in each wind-light output scene in the target wind-light output scene set based on a pre-constructed risk prevention and control model.

The risk prevention and control model is a main power distribution network risk prevention and control model which aims at minimum comprehensive operation risk indexes and minimum comprehensive operation cost.

In the embodiment, a mathematical optimization algorithm can be used for solving the optimal power flow problem, and the optimization algorithm adjusts the power values of the power supply and the load through iterative calculation so as to meet the objective function and the constraint condition, and evaluates and verifies the optimal power flow solving result; in some embodiments, a preset particle swarm algorithm may be used to solve an optimal power flow of the risk prevention and control model, so as to adjust energy storage and demand response of the main power distribution network under each wind-solar output scene according to the optimal power flow, so as to perform risk prevention and control on the main power distribution network; in other embodiments, the random power flow may be solved by using a semi-invariant and Gram-Charlie series expansion method, so as to obtain the voltage probability distribution of each node of the whole network, obtain the voltage probability density distribution of each node of the whole network, calculate the corresponding voltage out-of-limit risk probability and severity according to the voltage probability density, and further calculate the risk index value, thereby realizing the quantitative evaluation of risks.

acquiring power flow parameter data of a power distribution network in each wind-light power generation scene during operation;

Determining at least two power distribution network security risk indexes corresponding to the wind-light output scenes in the preset time granularity based on scene probabilities and tide parameter data corresponding to the wind-light output scenes respectively aiming at each wind-light output scene in the target wind-light output scene set;

and constructing a risk prevention and control model corresponding to the wind-light output scene according to the safety risk indexes of each power distribution network.

The power flow parameter data may be understood as parameter data contained in each node when the power distribution network operates, where the parameter data may include, but is not limited to: the method comprises the steps of (1) voltage amplitude of a node i in a power distribution network, and upper limit and lower limit of the voltage amplitude of the node i; active power sum of line l transmissionMaximum active power allowed to be transmitted for line l; the active load amount of node i and the active load value of node i.

In this embodiment, the power distribution network security risk indicator at least includes: node voltage out-of-limit risk indicators, line power out-of-limit risk indicators and system load loss risk indicators. For each wind-light output scene in the target wind-light output scene set, at least two power distribution network safety risk indexes corresponding to the wind-light output scene in the preset time granularity can be determined based on scene probabilities respectively corresponding to the wind-light output scenes and tide parameter data, and a risk prevention and control model corresponding to the wind-light output scene is constructed according to the safety risk indexes of each power distribution network. In this embodiment, different risk indexes may be formed by different tide parameter data, and specifically, a node voltage out-of-limit risk index may be constructed according to a voltage amplitude of a node i in the power distribution network, and an upper limit and a lower limit of the voltage amplitude of the node i; active power sum transmitted according to line lConstructing a line power out-of-limit risk index for the maximum active power allowed to be transmitted by the line l; according to the active load quantity of the node i and the active load value of the node i, a system load loss risk index can be constructed; the present embodiment is not limited herein.

In one embodiment, the power flow parameter data includes: under the conditions of the voltage amplitude of the node i and the upper limit and the lower limit of the voltage amplitude of the node i in the power distribution network, the node voltage out-of-limit risk index is expressed as follows by a formula: Wherein ρ (E _k) is the scene probability of the power distribution network node system state E _k; n _D is the total number of nodes of the power distribution network node system; v _i is the voltage amplitude of node i; /(I) V _i is the upper limit and the lower limit of the voltage amplitude of the node i respectively;

The tide parameter data includes: active power sum of line l transmission In the case of the maximum active power allowed to be transmitted by the line l, the line power out-of-limit risk indicator is expressed as: /(I)Wherein ρ (E _k) is the scene probability of the system state E _k; n _L is the total number of system lines; p _l is the active power transmitted by line l; /(I)Maximum active power allowed to be transmitted for line l;

the tide parameter data includes: under the conditions of the active load quantity of the node i and the active load value of the node i, the system load loss risk index is expressed as follows by a formula: Wherein ρ (E _k) is the scene probability of the system state E _k; p _cut,i is the active load amount of node i; p _d,i is the active load value of node i.

In this implementation, a risk index system of the main power distribution network risk prevention and control method of wind-solar output fluctuation is established, a short-term safety risk index system can be established from three angles of voltage, power and load, and an operation risk index system covering node voltage out-of-limit risk R _V, line active power out-of-limit risk R _L and system loss load R _C is established, specifically, node voltage out-of-limit risk indexes are expressed as follows by a formula: Wherein ρ (E _k) is the scene probability of the power distribution network node system state E _k; n _D is the total number of nodes of the power distribution network node system; v _i is the voltage amplitude of node i; /(I) V _i is the upper limit and the lower limit of the voltage amplitude of the node i respectively; the line power out-of-limit risk index is expressed as: Wherein ρ (E _k) is the scene probability of the system state E _k; n _L is the total number of system lines; p _l is the active power transmitted by line l; /(I) Maximum active power allowed to be transmitted for line l; the system load loss risk index is expressed as: /(I)Wherein ρ (E _k) is the scene probability of the system state E _k; p _cut,i is the active load amount of node i; p _d,i is the active load value of node i.

In one embodiment, the pre-built risk prevention and control model is formulated as:

Wherein, Expressed as minimum overall operational risk; /(I)Expressed as the minimum overall operating cost; r _V is a node voltage out-of-limit risk index; r _L is a line power out-of-limit risk index; r _C is a system load loss risk index; wherein ρ _t is the purchase price of electricity in different time periods; p _loss (t) is the net loss; s _Gi (t) is the cost of the controllable unit; z _Pi (t) is the interrupt cost; n _G is expressed as the number of controllable units; n _l represents the number of user interruptible loads;

Constraints of the risk prevention and control model include: demand response constraints, energy storage system (EnergyStorageSystem, ESS) constraints, fan constraints, photovoltaic constraints, generator constraints, reactive compensation device (STATICVAR COMPENSATOR, SVC) constraints, capacitor Binder (CB) constraints, node voltage constraints, and branch current constraints.

In this embodiment, both safety and economy are considered, a main power distribution network risk prevention and control model with minimum comprehensive operation risk index and minimum comprehensive operation cost as targets is established, and power output, node voltage and line transmission power of a generator corresponding to each system state and operation condition are obtained. These models are used to describe the topology and electrical parameters of the network; data acquisition and verification: real-time data of the power distribution network is collected, including line impedance, transformer parameters, load requirements, power supply states and the like. Through verification and correction of the data, the accuracy and reliability of the used data are ensured; the objective function definition: an objective function of the optimization problem is determined. The most common goal is to minimize power loss in the distribution network, and other factors such as minimizing voltage bias, load imbalance, etc. may also be considered. The selection of the objective function is determined according to the specific situation and the optimization target; constraint definition: constraints of the optimization problem are defined. These constraints include electrical constraints such as voltage, current, power factor, etc., and equipment operational constraints such as line load constraints, switching state constraints, etc. These constraints are used to ensure that the optimization solution meets the constraints of grid safety and equipment operation; solving an optimization algorithm: and solving the optimal power flow problem by using a mathematical optimization algorithm.

In one embodiment, the constraints due to the risk prevention model include: the method comprises the following steps of demand response constraint, energy storage system ESS constraint, fan constraint, photovoltaic constraint, generator constraint, SVC constraint, capacitor CB constraint, node voltage constraint and branch current constraint, wherein each constraint condition is described one by one; demand response constraints are formulated as: And/> Wherein/>The variable quantity generated by the response of the demand side at the time t of the n node is obtained; /(I)The load quantity before the response of the demand side at the moment t of the n node is; /(I)AndRespectively generating upper and lower limits of variation for the response of the demand side at the time t of the n node; kappa is the allowable variable range of electricity consumption in one period of the user; t is expressed as the number of hours of a day, and the value is 24; n _b represents the total number of nodes;

the charge and discharge state constraint in the energy storage system ESS constraint is expressed as: Wherein/> And/>Respectively charging and discharging states of the energy storage j at the moment t, and indicating 3 states of charging, discharging and non-charging and non-discharging when the constraint limit is within 1; n ^ES is the number of energy storage devices in the main distribution grid;

Fan constraint, formulated as: Wherein/> And/>Active power and reactive power sent by the fan j at the moment t are respectively; /(I)The maximum active power emitted by the fan j at the moment t; tan θ _WT is the fan power factor; n ^WT is the number of fans in the main power distribution network;

photovoltaic constraints, formulated as: Wherein/> And/>Active power and reactive power sent by the fan j at the moment t are respectively; /(I)The maximum active power emitted by the fan j at the moment t; tan θ _PV is the photovoltaic power factor; n ^PV is the number of fans in the main power distribution network;

Generator constraints, formulated as: Wherein/> Active power generated by the generator j at the moment t; /(I)Maximum active power generated by the generator j at the time t; n ^MT is the number of generators in the main distribution network;

capacitor CB constraint, formulated as: Wherein/> The reactive compensation quantity of the discrete reactive compensation device j at the time t is obtained; /(I)Switching the group number for the discrete reactive power compensation device j at the time t; /(I)Reactive compensation quantity switched for a single group of discrete reactive compensation devices j; /(I)The maximum switching group number of the discrete reactive power compensation device j; n ^CB is the number of discrete reactive power compensation devices in the main power distribution network;

Node voltage constraints, formulated as: wherein V _n,t is the voltage of the node n at the time t; /(I) And/>Upper and lower limits allowed for the voltage on node n;

branch current constraints, formulated as: Wherein/> And/>Upper and lower limits for the branch l to allow the flow of current; i _l,t is the value of the current flowing in the branch l at time t.

According to the technical scheme, the wind-light output scene set is determined through the obtained wind-light output prediction data, the wind-light output actual measurement data, the pre-added random noise and the pre-trained WGAN-GP model, scene reduction is carried out on the wind-light output scene set according to a preset clustering algorithm, the target wind-light output scene set after scene reduction and the corresponding scene probability thereof are obtained, scene reduction can be carried out, typical characteristics of wind-light output can be better described by the reduced scene, and the fluctuation of wind-light output is reflected; and performing risk prevention and control on the main power distribution network in each wind-light output scene in the target wind-light output scene set through a pre-constructed risk prevention and control model, so that the condition that the traditional wind-light output scene set is regulated only by a controllable unit can be improved, and the running risk of the main power distribution network can be effectively reduced.

In an embodiment, fig. 3 is a flowchart of another risk prevention and control method for a main distribution network according to an embodiment of the present invention, where, based on the foregoing embodiments, a wind-light output scene set is determined according to wind-light output prediction data, wind-light output actual measurement data, pre-added random noise, and a pre-trained WGAN-GP model; scene reduction is carried out on the scene set of the wind and light output according to a preset clustering algorithm, and a target scene set of the wind and light output after scene reduction and a corresponding scene probability are obtained; and performing risk prevention and control on the main power distribution network in each wind-light output scene in the target wind-light output scene set based on a pre-constructed risk prevention and control model, so as to further refine the risk prevention and control.

As shown in fig. 3, the risk prevention and control method for the main distribution network in this embodiment may specifically include the following steps:

S310, wind and light output prediction data and wind and light output actual measurement data in a preset time period are obtained.

S320, inputting wind-light output prediction data and at least two groups of random noise into a generator in a pre-trained WGAN-GP model to obtain generation data, and inputting the generation data into a discriminator in the pre-trained WGAN-GP model.

Wherein the random noise is in accordance with a normal distribution.

In this embodiment, wind-solar output prediction data and at least two sets of random noise are input into a generator in a pre-trained WGAN-GP model to obtain generation data, and the generation data is input into a discriminator in the pre-trained WGAN-GP model. The extraction generator respectively inputs wind power, photovoltaic predicted values and 200 groups of noise, and the noise and the predicted values are spliced and then input into the generator to generate a photovoltaic wind power output scene set based on the predicted values.

S330, inputting the actual measurement data of the wind and light output into a discriminator in a pre-trained WGAN-GP model.

In the embodiment, the measured data of wind and light output are input into a discriminator in a pre-trained WGAN-GP model.

S340, outputting at least two wind and light output scenes from the discriminator, and forming a wind and light output scene set from the at least two wind and light output scenes.

In the embodiment, at least two wind-light output scenes are output from a discriminator in a WGAN-GP model, and the wind-light output scenes form a wind-light output scene set.

S350, selecting a preset number of wind and light output scenes from the wind and light output scene set as an initial clustering center.

The initial cluster centers are K cluster centers which are determined initially, and the initial cluster centers can be set in a self-defined mode according to requirements manually.

In this embodiment, a preset number of wind-light output scenes are selected from the wind-light output scene set as an initial clustering center, which can be understood that, according to the requirement of scene reduction, an appropriate K value, that is, the number of final scenes to be reduced, is selected, and K samples are randomly selected from the data set as the initial clustering center.

S360, determining a first distance from other wind and light output scenes in the wind and light output scene set to an initial clustering center by adopting a preset similarity measurement method, and dividing the other wind and light output scenes to the closest clustering center according to the first distance to form a first clustering center.

The preset similarity measurement method may include one of euclidean distance and cosine similarity calculation method. The first distance may be understood as a distance from the initial cluster center of other wind-light output scenes than the wind-light output scene selected as the initial cluster center, which distance may characterize the similarity.

In this embodiment, a preset similarity measurement method is adopted to determine a first distance from other wind-light output scenes in the wind-light output scene set to the initial clustering center, and the other wind-light output scenes are divided into the closest clustering centers according to the first distance to form the first clustering center.

S370, for each first clustering center, respectively calculating a second distance of the wind-light output scenes in each first clustering center, updating the first clustering centers according to the second distance to obtain new clustering centers, and returning to the step of dividing other wind-light output scenes to the closest clustering center according to the first distance to form the first clustering centers until the first clustering centers are not changed any more or the preset iteration times are reached, so as to obtain a target clustering result.

Wherein the second distance characterizes a total distance of the wind-light output scenes in the first cluster center. The new cluster center is the new cluster center obtained by updating the first cluster center; and the target clustering result is the final form clustering result.

In this embodiment, the target clustering result includes: the number of samples or sample weights in each cluster; and the scene probability sum of all the wind-light power scenes is 1.

In this embodiment, for each first cluster center, a second distance of the wind-light output scene in each first cluster center is calculated respectively, the first cluster center is updated according to the second distance to obtain a new cluster center, and the step of dividing other wind-light output scenes to the cluster center closest to the first cluster center according to the first distance is returned to form the first cluster center until the first cluster center is not changed any more or a preset iteration number is reached, so as to obtain a target cluster result.

S380, determining a reduced target wind-light output scene set according to the target clustering result.

The target clustering result comprises the following steps: the number of samples or sample weights included in each cluster; and the scene probability of each wind-light power scene in the target wind-light power scene set corresponds to the scene probability, and the scene probability sum of all the wind-light power scenes is 1.

In this embodiment, the probability of each cluster may be calculated from the cluster allocation result of each data point. For example, the number of samples or sample weights in each cluster may be used as an estimate of the probability, and a reduced set of scenes and corresponding probability weights may be output as desired.

For better understanding, the scene reduction by using the K-means clustering method is performed to obtain a reduced scene set, and fig. 4 is a reduced 5 typical wind power output scene graph according to an embodiment of the present invention; FIG. 5 is a graph of a scene of reduced 5 typical photovoltaic output according to an embodiment of the present invention; in this embodiment, a certain daily wind power, a photovoltaic prediction value and an actual measurement value are selected as samples, and the samples are not trained and can verify the generalization of the model. The extraction generator respectively inputs wind power, photovoltaic predicted values and 200 groups of noise, the noise and the predicted values are spliced and then input into the generator to generate a photovoltaic wind power output scene set based on the predicted values, on the basis of 200 generated wind power typical scenes, the scene is reduced by adopting a K-medoids clustering method, the reduced scenes can better describe typical characteristics of wind power output, the fluctuation of wind power output is reflected, and the generated wind power output typical scenes are shown in figures 4 and 5.

S390, solving the optimal power flow of the risk prevention and control model by adopting a preset particle swarm algorithm.

In this embodiment, the preset particle swarm algorithm is a particle optimization algorithm, and may be preset to solve the optimal power flow of the risk prevention and control model, specifically, initialize the particle swarm: the position and velocity of a population of particles (representing a possible solution) are randomly initialized. The position of each particle represents a set of solutions for the parameters node voltage and generator output in the power system, and the velocity represents the direction and velocity of the particle search in the solution space. Updating particle position and velocity: and calculating new positions and speeds by updating formulas according to the current positions and speeds of the particle swarms. Updating the formula typically includes taking into account the individual and population optimal portions, i.e., the particles update position and velocity based on their own historical optimal solutions and the overall population's historical optimal solutions. Evaluating an objective function: for each new position of the particle, the value of the objective function, i.e. the fitness of the optimal power flow problem, is calculated. Updating individual and population optima: and updating the individual optimal solution of each particle according to the new fitness value, and selecting the optimal solution as a group optimal solution. Judging a stopping condition: it is checked whether the stop condition is satisfied. The stopping condition may be reaching a maximum number of iterations, the objective function value converging to a certain threshold or meeting a constraint, etc. Terminating or continuing the iteration: if the stop condition is met, the algorithm is terminated and the optimal solution is returned. Otherwise, the particle position and velocity continue to be updated.

In an embodiment, solving the optimal power flow of the risk prevention and control model by adopting a preset particle swarm algorithm includes:

Judging whether the group optimal solution meets the stopping condition, and if so, returning to the optimal solution; if not, returning to the step of updating the current position according to the first historical optimal solution of the particles and the second historical optimal solution of the whole group to obtain the updated new position, and continuing to update the position and the speed of the particles until the stopping condition is met.

Wherein the stop condition includes one of: the objective function value converges to a preset threshold value, the objective function value meets constraint conditions, and the preset iteration times are reached.

In this embodiment, the current position and current speed of the initial population of particles are randomly initialized; wherein the initial population of particles represents a possible solution; the position of each particle represents a set of solutions to the node voltage and generator output parameters in the power system, and the velocity of each particle represents the direction and velocity of the particle searching in the solution space; updating the current position according to the first historical optimal solution of the particles and the second historical optimal solution of the whole population to obtain an updated new position; calculating an output value of the risk prevention and control model according to the new position; the output value represents the fitness of the optimal power flow problem; updating the individual optimal solution of each particle according to the fitness, and selecting the optimal solution as a group optimal solution; judging whether the group optimal solution meets the stopping condition, and if so, returning to the optimal solution; if not, returning to the step of updating the current position according to the first historical optimal solution of the particles and the second historical optimal solution of the whole group to obtain the updated new position, and continuing to update the position and the speed of the particles until the stopping condition is met.

S3100, adjusting energy storage and demand response of the main distribution network under each wind-light output scene according to the optimal power flow so as to perform risk prevention and control on the main distribution network.

In this embodiment, energy storage and demand response of the main power distribution network in each scene of wind-light output are adjusted according to the optimal power flow so as to perform risk prevention and control on the main power distribution network, which can be understood that according to the optimal power flow result, the operation strategy of the power distribution network, such as load adjustment, operation sequence change, switch state adjustment, etc., is adjusted, in this embodiment, the energy storage and demand response are mainly performed.

For example, in order to better understand the operation conditions of the demand response and the energy storage, that is, the operation conditions of the demand response and the energy storage participate in the coordinated control change trend chart, fig. 6 is a power change trend chart of the interruptible load 24h according to an embodiment of the present invention; FIG. 7 is a graph showing a trend of power variation of a load 24h with time shift according to an embodiment of the present invention; fig. 8 is a graph showing a trend of the electric quantity change of the energy storage device 24h according to an embodiment of the present invention.

According to the technical scheme, wind-light output prediction data and at least two groups of random noise are input into a generator in a pre-trained WGAN-GP model to obtain generation data, and the generation data are input into a discriminator in the pre-trained WGAN-GP model; inputting wind-light output actual measurement data into a discriminator in a pre-trained WGAN-GP model, outputting from the discriminator to obtain at least two wind-light output scenes, and forming a wind-light output scene set from the at least two wind-light output scenes, so that dynamic information of a wind-light output curve can be extracted better, and the quality of a wind-light output generated sample is higher; determining first distances from other wind-light output scenes in the wind-light output scene set to an initial clustering center by adopting a preset similarity measurement method, dividing the other wind-light output scenes to the clustering center closest to the initial clustering center according to the first distances to form first clustering centers, respectively calculating second distances of the wind-light output scenes in each first clustering center for each first clustering center, updating the first clustering center according to the second distances to obtain a new clustering center until the first clustering center is not changed any more or the preset iteration times are reached to obtain a target clustering result, determining a reduced scene set and corresponding probability weights thereof according to the target clustering result, performing scene reduction, and performing scene reduction to better describe typical characteristics of wind-light output, so that the fluctuation of the wind-light is reflected; the optimal power flow of the risk prevention and control model is solved by adopting a preset particle swarm algorithm, and the energy storage and demand response of the main distribution network under each wind-light output scene are adjusted according to the optimal power flow so as to prevent and control the risk of the main distribution network, so that the condition that the traditional method only depends on the adjustment of a controllable unit is further improved, and the running risk of the main distribution network can be effectively reduced.

Exemplary, to facilitate better understanding of the topology of the main distribution network for risk prevention and control. Fig. 9 is a topological structure diagram of risk prevention and control for a main power distribution network according to an embodiment of the present invention, where as shown in fig. 9, a main network portion selects an IEEE30 node system, and connects nodes 8 and 14 to the power distribution network, the power distribution network structure adopts an IEEE33 node system, and connects nodes 13, 16, 19, 23 and 31 of the power distribution network to diesel engines, gas turbines, energy storage devices, photovoltaics and wind turbines, and the algorithm is shown in fig. 6. The reference voltage of the power distribution network part example is 12.66kV, and the reference power is 10MW. Conventional load expected peak summation takes 4.125MW. The power factor of each load and power supply was 0.95. In this scenario, the main distribution network voltage out-of-limit risk value, the power flow out-of-limit risk value and the load loss risk value can be obtained as shown in the table. The condition 1 is to consider the influence of the fluctuation of the wind and light output on the running risk index of the main power distribution network, and the condition 2 is to not consider the influence of the fluctuation of the wind and light output on the running risk index of the main power distribution network. It can be seen from table 1 that the fluctuation of the wind-solar power output also has a larger influence on the running risk of the main distribution network, so that reasonable measures are necessary to reduce the running risk of the main distribution network.

Table 1 considers the operational risk index of the main distribution network under the wind-solar fluctuation

And considering as the schedulable resource of load side, flexible electric load possesses the demand response characteristic, through guiding flexible electric load to participate in main distribution network coordination optimization control, improve the tradition and only rely on the condition that controllable unit adjusted, can effectively reduce main distribution network operation risk, specific case is shown as table 2. The case 1 is that energy storage and demand response participate in coordination optimization control; the case 2 is that no energy storage and demand response participate in coordination optimization control; the condition 3 is that only the demand response participates in coordination optimization control; case 4 is that only energy storage participates in coordinated optimization control. According to the table 2, the flexible electric load is guided to participate in the coordination optimization control of the main distribution network, so that the risk of losing the load, the risk of out-of-limit voltage and the risk of out-of-limit power flow of the main distribution network are reduced to a certain extent, and the effectiveness of the provided risk prevention and control method is verified.

Table 2 considers operational risk indicators for a main distribution grid under flexible electrical loads

In an embodiment, fig. 10 is a block diagram of a risk prevention and control device for a main distribution network according to an embodiment of the present invention, where the device is applicable to a situation when risk prevention and control is performed on a main distribution network in an electric power internet of things, and the device may be implemented by hardware/software. The risk prevention and control processing method for the main distribution network can be configured in electronic equipment to realize the risk prevention and control processing method for the main distribution network in the embodiment of the invention.

As shown in fig. 10, the apparatus includes: an acquisition module 1010, a determination module 1020, a reduction module 1030, and a prevention and control module 1040;

the acquiring module 1010 is configured to acquire wind-light output prediction data and wind-light output actual measurement data within a preset period of time;

The determining module 1020 is configured to determine a wind-light output scene set according to the wind-light output prediction data, the wind-light output actual measurement data, the pre-added random noise, and a pre-trained generation countermeasure network WGAN-GP model with gradient penalty;

the reduction module 1030 is configured to perform scene reduction on the wind and light output scene set according to a preset clustering algorithm, so as to obtain a target wind and light output scene set after scene reduction and a corresponding scene probability thereof;

The prevention and control module 1040 is configured to perform risk prevention and control on the main power distribution network in each wind-light output scene in the target wind-light output scene set based on a pre-constructed risk prevention and control model; the risk prevention and control model is a main power distribution network risk prevention and control model which aims at minimum comprehensive operation risk indexes and minimum comprehensive operation cost.

According to the embodiment of the invention, a determining module determines a wind-light output scene set through the obtained wind-light output prediction data, the wind-light output actual measurement data, the pre-added random noise and a pre-trained WGAN-GP model, and a reducing module reduces the wind-light output scene set according to a preset clustering algorithm to obtain a target wind-light output scene set after scene reduction, so that the scene reduction can be carried out, the reduced scene can better describe typical characteristics of wind-light output, and the fluctuation of wind-light output is reflected; and the prevention and control module is used for performing risk prevention and control on the main distribution network in each wind-light output scene in the target wind-light output scene set through a pre-constructed risk prevention and control model, so that the condition that the traditional device is only regulated by a controllable unit can be improved, and the running risk of the main distribution network can be effectively reduced.

In one embodiment, the determining module 1020 further comprises:

The first input unit is used for inputting the wind-solar output prediction data and at least two groups of random noise into a generator in a pre-trained WGAN-GP model to obtain generation data, and inputting the generation data into a discriminator in the pre-trained WGAN-GP model; wherein the random noise is in accordance with normal distribution;

the second input unit is used for inputting the wind-solar output actual measurement data into a discriminator in a pre-trained WGAN-GP model;

The scene set determining unit is used for outputting at least two wind and light output scenes from the discriminator and forming the wind and light output scenes into a wind and light output scene set.

In one embodiment, the WGAN-GP model training process includes:

A gradient penalty term is added in the loss function of the discriminator, the gradient penalty term has a weight coefficient, a preset deep learning optimization algorithm is adopted to update the weight coefficient respectively, and the step of acquiring the historical wind-solar output data of a preset area as a sample set is returned until the difference between the generated sample output by the discriminator and the actual measurement value of the historical wind-solar output is minimum;

wherein, the objective function in the discriminator is expressed as: Wherein E (·) represents the expected value; d (G (z)) is a probability that the generated data G (z) is discriminated as true in the discriminator; d (x) represents the probability that the true data x is discriminated as true in the discriminator; the distribution of the real data x is x-p _data(x); z represents the random noise, wherein the distribution of the random noise is z-P _z;P_X which is the real distribution of wind and light data x; x' =εx+ (1- ε) G (z); epsilon represents a random number; λ represents a weight coefficient of the gradient penalty term; the expression of the norm is a mathematical calculation method; denoted hamiltonian. /(I)

In one embodiment, the reduction module 1030 further includes:

the selecting unit is used for selecting a preset number of wind-light output scenes from the wind-light output scene set to serve as an initial clustering center;

The cluster center determining unit is used for determining a first distance from other wind-light output scenes in the wind-light output scene set to the initial cluster center by adopting a preset similarity measurement method, and dividing the other wind-light output scenes to the cluster center closest to the first distance according to the first distance so as to form a first cluster center; wherein the first distance characterizes similarity;

The result determining unit is used for respectively calculating a second distance of the wind-light output scenes in each first clustering center, updating the first clustering centers according to the second distance to obtain new clustering centers, and returning to the step of dividing the other wind-light output scenes to the closest clustering centers according to the first distance to form the first clustering centers until the first clustering centers are not changed any more or reach preset iteration times to obtain target clustering results; wherein the second distance characterizes a total distance of the wind-light output scenes in the first cluster center;

the scene set determining unit is used for determining a reduced target wind-light output scene set according to the target clustering result;

The tide parameter data comprises: active power sum of line l transmission In the case of the maximum active power allowed to be transmitted by the line l, the line power out-of-limit risk indicator is expressed as: /(I)Wherein ρ (E _k) is the scene probability of the system state E _k; n _L is the total number of system lines; p _l is the active power transmitted by line l; /(I)Maximum active power allowed to be transmitted for line l; /(I)

In one embodiment, the prevention and control module 1040 includes:

the power flow determining unit is used for solving the optimal power flow of the risk prevention and control model by adopting a preset particle swarm algorithm;

and the prevention and control unit is used for adjusting the energy storage and demand response of the main distribution network in each wind-light output scene according to the optimal power flow so as to prevent and control the risk of the main distribution network.

In an embodiment, the power flow determination unit comprises:

An initialization subunit, configured to randomly initialize a current position and a current speed of the initial group of particles; wherein the initial population of particles represents a possible solution; the position of each particle represents a set of solutions to the node voltage and generator output parameters in the power system, and the velocity of each particle represents the direction and velocity of the particle searching in the solution space;

The updating subunit is used for updating the current position according to the first historical optimal solution of the particles and the second historical optimal solution of the whole population to obtain an updated new position;

an output value determining subunit, configured to calculate an output value of the risk prevention and control model according to the new position; the output value represents the fitness of the optimal power flow problem;

the optimal solution determining subunit is used for updating the individual optimal solution of each particle according to the fitness and selecting the optimal solution as a group optimal solution;

Judging subunit, judging whether the group optimal solution meets the stopping condition, if so, returning to the optimal solution; if not, returning to the step of updating the current position according to the first historical optimal solution of the particles and the second historical optimal solution of the whole population to obtain an updated new position, and continuing to update the position and the speed of the particles until the stopping condition is met; wherein the stop condition includes one of: the objective function value converges to a preset threshold value, the objective function value meets constraint conditions, and the preset iteration times are reached.

The risk prevention and control processing device for the main distribution network provided by the embodiment of the invention can execute the risk prevention and control processing method for the main distribution network applied to the financial system provided by any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

In an embodiment, fig. 11 is a schematic structural diagram of an electronic device according to an embodiment of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. Electronic equipment may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the inventions described and/or claimed herein.

As shown in fig. 11, the electronic device 10 includes at least one processor 11, and a memory such as a Read Only Memory (ROM) 12, a Random Access Memory (RAM) 13, etc., communicatively connected to the at least one processor 11, wherein the memory stores a computer program executable by the at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the Read Only Memory (ROM) 12 or the computer program loaded from the storage unit 18 into the Random Access Memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 may also be stored. The processor 11, the ROM 12 and the RAM 13 are connected to each other via a bus 14. An input/output (I/O) interface 15 is also connected to bus 14.

Various components in the electronic device 10 are connected to the I/O interface 15, including: an input unit 16 such as a keyboard, a mouse, etc.; an output unit 17 such as various types of displays, speakers, and the like; a storage unit 18 such as a magnetic disk, an optical disk, or the like; and a communication unit 19 such as a network card, modem, wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.

The processor 11 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various processors running machine learning model algorithms, digital Signal Processors (DSPs), and any suitable processor, controller, microcontroller, etc. The processor 11 performs the various methods and processes described above, such as the risk prevention and control method of the main distribution network.

In some embodiments, the risk prevention and control processing method of the main distribution network may be implemented as a computer program, which is tangibly embodied on a computer readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 10 via the ROM 12 and/or the communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the risk prevention and control method of the main distribution network described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the risk prevention method of the main distribution network in any other suitable way (e.g. by means of firmware).

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.

A computer program for carrying out methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general purpose computer, special purpose computer, or other programmable risk prevention and control device for a main distribution network, such that the computer programs, when executed by the processor, cause the functions/operations specified in the flowchart and/or block diagram to be performed. The computer program may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of the present invention, a computer-readable storage medium may be a tangible medium that can contain, or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer readable storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, the computer readable storage medium may be a machine readable signal medium. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which a user can provide input to the electronic device. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), blockchain networks, and the internet.

The computing system may include clients and servers. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical hosts and VPS service are overcome.

It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps described in the present invention may be performed in parallel, sequentially, or in a different order, so long as the desired results of the technical solution of the present invention are achieved, and the present invention is not limited herein.

The above embodiments do not limit the scope of the present invention. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of the present invention.

Claims

1. The risk prevention and control method for the main distribution network is characterized by comprising the following steps of:

2. The method of claim 1, wherein said determining a scene set of wind and light output from said wind and light output prediction data, said wind and light output measured data, pre-added random noise, and a pre-trained WGAN-GP model, further comprises:

3. The method according to any one of claims 1 or 2, wherein the training process of the WGAN-GP model comprises:

4. A method according to claim 3, wherein the objective function in the arbiter is formulated as: wherein E (·) represents the expected value; d (G (z)) is a probability that the generated data G (z) is discriminated as true in the discriminator; d (x) represents the probability that the true data x is discriminated as true in the discriminator; the distribution of the real data x is x-p _data(x); z represents the random noise, wherein the distribution of the random noise is z-P _z;P_X which is the real distribution of wind and light data x; x' =εx+ (1- ε) G (z); epsilon represents a random number; λ represents a weight coefficient of the gradient penalty term; the expression of the norm is a mathematical calculation method; denoted hamiltonian.

5. The method of claim 1, wherein the scene reducing the scene set of the wind-light output according to the preset clustering algorithm to obtain a scene reduced target scene set of the wind-light output, further comprising:

6. The method of claim 1, wherein the constructing of the risk prevention model further comprises:

7. The method of claim 6, wherein the power flow parameter data comprises: under the conditions of the voltage amplitude of the node i and the upper limit and the lower limit of the voltage amplitude of the node i in the power distribution network, the node voltage out-of-limit risk index is expressed as follows: Wherein ρ (E _k) is the scene probability of the power distribution network node system state E _k; n _D is the total number of nodes of the power distribution network node system; v _i is the voltage amplitude of node i; /(I) V _i is the upper limit and the lower limit of the voltage amplitude of the node i respectively;

The tide parameter data comprises: active power sum of line l transmission In the case of the maximum active power allowed to be transmitted by the line l, the line power out-of-limit risk indicator is expressed as: /(I)Wherein ρ (E _k) is the scene probability of the system state E _k; n _L is the total number of system lines; p _l is the active power transmitted by line l; /(I)Maximum active power allowed to be transmitted for line l;

8. The method of any one of claims 1 or 7, wherein the pre-constructed risk prevention and control model is formulated as:

9. The method of claim 8, wherein the demand response constraint is formulated as: And/> Wherein/>The variable quantity generated by the response of the demand side at the time t of the n node is obtained; /(I)The load quantity before the response of the demand side at the moment t of the n node is; /(I)AndRespectively generating upper and lower limits of variation for the response of the demand side at the time t of the n node; kappa is the allowable variable range of electricity consumption in one period of the user; t is expressed as the number of hours of a day, and the value is 24; n _b represents the total number of nodes;

The SVC constraint of the reactive power compensation device is expressed as: Wherein, The reactive compensation quantity of the continuous reactive compensation device j at the time t is obtained; /(I)And/>Reactive compensation upper and lower limits of the continuous reactive compensation device j; n ^SVC is the number of continuous reactive power compensation devices in the main power distribution network;

The capacitor CB constraint is formulated as: Wherein/> The reactive compensation quantity of the discrete reactive compensation device j at the time t is obtained; /(I)Switching the group number for the discrete reactive power compensation device j at the time t; /(I)Reactive compensation quantity switched for a single group of discrete reactive compensation devices j; /(I)The maximum switching group number of the discrete reactive power compensation device j; n ^CB is the number of discrete reactive power compensation devices in the main power distribution network;

10. The method according to claim 1, wherein performing risk prevention and control on the main distribution network in each wind-light output scene in the target wind-light output scene set based on a pre-constructed risk prevention and control model comprises:

11. The method of claim 10, wherein solving the optimal power flow of the risk prevention and control model using a preset particle swarm algorithm comprises:

12. The utility model provides a risk prevention and control device of main distribution network which characterized in that includes:

The determining module is used for determining a wind and light output scene set according to the wind and light output prediction data, the wind and light output actual measurement data, the pre-added random noise and a pre-trained generation countermeasure network WGAN-GP model with gradient penalty;

The prevention and control module is used for performing risk prevention and control on the main distribution network in each wind-light output scene in the target wind-light output scene set based on a pre-constructed risk prevention and control model; the risk prevention and control model is a main power distribution network risk prevention and control model which aims at minimum comprehensive operation risk indexes and minimum comprehensive operation cost.

13. The apparatus of claim 12, wherein the determining module further comprises:

14. The apparatus of any one of claims 12 or 13, wherein the training process of the WGAN-GP model comprises:

15. The apparatus of any of claims 13, wherein the objective function in the arbiter is formulated as:

Wherein E (·) represents the expected value; d (G (z)) is a probability that the generated data G (z) is discriminated as true in the discriminator; d (x) represents the probability that the true data x is discriminated as true in the discriminator; the distribution of the real data x is x-p _data(x); z represents the random noise, wherein the distribution of the random noise is z-P _z;P_X which is the real distribution of wind and light data x; x' =εx+ (1- ε) G (z); epsilon represents a random number; λ represents a weight coefficient of the gradient penalty term; the expression of the norm is a mathematical calculation method; Represented as hamiltonian.

16. The apparatus of claim 12, wherein the downscaling module further comprises:

17. The apparatus of claim 12, wherein the construction of the risk prevention model further comprises:

18. The apparatus of claim 17, wherein the power flow parameter data comprises: under the conditions of the voltage amplitude of the node i and the upper limit and the lower limit of the voltage amplitude of the node i in the power distribution network, the node voltage out-of-limit risk index is expressed as follows: Wherein ρ (E _k) is the scene probability of the power distribution network node system state E _k; n _D is the total number of nodes of the power distribution network node system; v _i is the voltage amplitude of node i; /(I) V _i is the upper limit and the lower limit of the voltage amplitude of the node i respectively;

the tide parameter data comprises: under the conditions of the active load quantity of the node i and the active load value of the node i, the system load loss risk index is expressed as follows by a formula: Wherein ρ (E _k) is the scene probability of the system state E _k; p _cut, i is the active load of node i; p _d,i is the active load value of node i.

19. The apparatus of any one of claims 12 or 18, wherein the pre-constructed risk prevention and control model is formulated as:

20. The apparatus of claim 19, wherein the demand response constraint is formulated as: And/> Wherein/>The variable quantity generated by the response of the demand side at the time t of the n node is obtained; /(I)The load quantity before the response of the demand side at the moment t of the n node is; And/> Respectively generating upper and lower limits of variation for the response of the demand side at the time t of the n node; kappa is the allowable variable range of electricity consumption in one period of the user; t is expressed as the number of hours of a day, and the value is 24; n _b represents the total number of nodes;

21. The apparatus of claim 12, wherein the prevention and control module further comprises:

22. The apparatus of claim 21, wherein the power flow determination unit comprises:

The judging subunit is used for judging whether the group optimal solution meets the stopping condition, and if so, returning to the optimal solution; if not, returning to the step of updating the current position according to the first historical optimal solution of the particles and the second historical optimal solution of the whole population to obtain an updated new position, and continuing to update the position and the speed of the particles until the stopping condition is met; wherein the stop condition includes one of: the objective function value converges to a preset threshold value, the objective function value meets constraint conditions, and the preset iteration times are reached.

23. An electronic device, the electronic device comprising:

At least one processor; and

A memory communicatively coupled to the at least one processor; wherein,

The memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the risk prevention method of the main distribution network of any one of claims 1-11.

24. A computer readable storage medium, characterized in that the computer readable storage medium stores computer instructions for causing a processor to implement the risk prevention and control method of a main distribution network according to any one of claims 1-11 when executed.